Description:

         A disjunctive interrogative has two different semantic interpretations: as an alternative (Alt.) question, or yes-no (Y/N) question, dependent on prosody.  These interpretations also differ in the types of licit answers: Alt. questions demand ‘exactly one’ of the listed alternatives, while Y/N questions can have many answers, including Yes, No, Both, Neither, often in combination with any of the disjuncts proposed. Semantic theories that model disjunctive questions and their licit answers rely on many assumptions.  First, the models are framed according to assumed canonical prosodic contours, which have been untested in the experimental field.  Certain responses are deemed ‘licit’ or ‘illicit’ without regard to syntactic variation or intonational differences.  A disjunctive question is assumed unbiased without prosody, but it is assumed that inserting either forces the Y/N interpretation.  The relationship between Y/N questions and Alt. questions is assumed to be a superset/subset relationship.

            Four experiments were conducted in order to examine the underlying assumptions of disjunctive questions and their licit responses.  A production experiment revealed the true restrictions on Alt. question contours; Alt. questions do not need a phrase break between disjuncts, but do rely on a contour in which the first disjunct ends in a relatively high pitch range, while the second disjunct (or full utterance) ends in a relatively low pitch range. 

           The text experiment exposed the inherent bias of the ambiguous string: a disjunctive question is biased to the Y/N interpretation when prosody is absent.  Furthermore, licit responses to a disjunctive question lie on a continuous scale, rather than a categorical one.  If the response is valid for one of the two interpretations, then participants always seem to rate the response as at least slightly acceptable.

            The perception experiment indicated that either can be inserted into either type of question and the meaning of the question will be derived from the intonation.  The production experiment results support this claim, as participants voluntarily inserted either into both Alt. and Y/N productions, when not given a specific utterance prompt.

            The artificial language experiment tested participants against different language conditions of the Alt. and Y/N question dichotomy: some conditions had two separate lexemes for the “or” in the question, while others had one; some conditions had monosyndetic constructions (the “or” occurred between disjuncts) while others had bisyndetic constructions (the “or” occurred before both disjuncts), or even a juxtaposition construction (no word for “or”).  Results indicated that participants think of Alt. questions as a specific subset of Y/N questions; the text experiment supports this conclusion as well.

           The data revealed shortcomings in frameworks intending to model interrogatives and their licit responses.  For example, Inquisitive Semantics relies on prosodic focus to distinguish between Alt. questions and Y/N questions. The production experiment revealed both types of questions can be produced with and without prosodic focus on both disjuncts.  Commitment Space Discourse, however, models Alt. questions and Y/N questions the same way; it lacks a method to ensure the ‘exactly one’ stipulation on Alt. responses is upheld.  Both frameworks required modifications in light of the experimental results.  Various mechanisms within each framework were explored in order to amend their shortcomings; however neither can fully account for the data from the experiments.